"As economists put a price tag on sky-high body mass indexes," Reuters reports, "policymakers as well as the private sector are mobilizing to find solutions to the obesity epidemic." The story lumps together all sorts of obesity-related costs, without regard to who pays them or why. It worries that people have to buy more gasoline to transport their bloated bodies, for instance, and it says the lower wages earned by obese women amount to "a $76 weekly obesity tax," although these costs are borne by fat people themselves and therefore hardly count as externalities justifying government intervention. Reuters likewise counts the cost to employers of obesity-related absenteeism and "presenteeism" (reduced productivity), although businesses can take such costs into account when deciding whom to employ and how much to pay them (which could help explain the relatively low wages of obese workers). But mainly Reuters worries about obesity-related medical costs, citing a recent study that puts them at $190 billion a year in the United States, more than twice as high as previously estimated.

That study, reported in the January issue of the Journal of Health Economics, is based on the same data source as the earlier estimate (which comes from research led by Duke University health economist Eric Finkelstein): the Medical Expenditure Panel Survey. But instead of the body-mass index (BMI) reported for each adult, the authors, Cornell economists John Cawley and Chad Meyerhoefer, used the BMI of the subject's oldest biological child, on the theory that respondents would be less apt to understate their children's weights than their own (or their spouses). Since "there is a strong genetic component to weight," Cawley and Meyerhoefer reasoned, children's BMIs should be a good indicator of their parents' BMIs. They also hoped this indirect measure would control for the possibility that people gain weight "after suffering an injury or chronic depression," which would lead to a noncausal association between obesity and health care spending. Cawley and Meyerhoefer say their approach identifies "the causal effect of obesity," as opposed to earlier studies, which gave us only "the association of obesity with medical care costs." 

As they later acknowledge, that is not quite true. In theory, using children's BMIs as proxies for their parents' BMIs allows Cawley and Meyerhoefer to measure the association between a genetic predispositon to obesity and medical costs. They argue that "the preponderance of evidence" indicates the impact of a "shared family environment" is "so small as to be undetectable and ignorable." But even assuming that's true, the authors concede, a predisposition to obesity may be associated with other genetic factors that boost medical expenses. In fact, if the predisposition to obesity is really a predisposition to dislike physical exertion and eat large amounts of high-calorie, low-nutrient food, lack of exercise and poor diet could be the main causes of ill health, as opposed to extra pounds per se. 

A closer look at Cawley and Meyerhoefer's data casts further doubt on the importance of government-recommended BMIs. Unlike earlier studies, their cost estimate compares the obese to the nonobese, included people deemed "overweight," whose medical costs are about the same as those of "healthy weight" (and who actually have lower mortality rates than people in the "healthy" range). For men and women together, the researchers note, "expenditures fall with BMI through the underweight and healthy weight categories, are relatively constant with BMI in the overweight category, then rise sharply with BMI through the obese category." 

Looking just at men, "expenditures drop sharply with BMI through the healthy weight category, fall modestly with BMI in the overweight category, then rise slowly at first then sharply with BMI in the obese category." In other words, higher BMIs are associated with lower health care costs among men, even after they pass a BMI of 25 and are officially considered "overweight." In fact, "the BMI value associated with minimum expenditures for men is roughly 30," the dividing line between "overweight" and "obese." For women, the picture is different:  

For a wide range of BMI, roughly 15–25, expected expenditures are relatively constant. As BMI rises beyond 25, however, expected expenditures rise through the overweight and obese categories, increasing rapidly at the high end of the obese range.

The difference in health care spending between obese and nonobese women is statistically significant, but the difference between obese and nonobese men is not. These sex differences may have to do with the weakness of BMI as a measure of obesity, since men are more likely to build extra muscle, which can make a fit person officially overweight or obese by raising his BMI.

For both sexes, Cawley and Meyerhoefer say, "the large average effects of obesity...are explained by relatively few individuals with very high BMI that incur very high medical expenditures." This pattern, which is similar to what Finkelstein has found, suggests that the merely portly and even the mildly obese should not be blamed for racking up outrageously high medical bills, because they don't. The vast majority of people who fail to heed the government’s BMI recommendations apparently are not contributing much to obesity-related health care spending.

What about the folks "with very high BMI"? I would say insurers should be free to charge them higher premiums, although the logic of ObamaCare's ban on discrimination against people with "pre-existing conditions" suggests otherwise. Shouldn't insurers be forced to charge obese people the same rates as thinner policyholders, especially given the "strong genetic component to weight"? Not quite, it turns out. Reuters notes that the Patient Protection and Affordable Care Act "allows employers to charge obese workers 30 percent to 50 percent more for health insurance if they decline to participate in a qualified wellness program," although that does not necessarily mean they will lose any weight.

As for the government's share of obesity-related medical costs, this study does not address the question of whether obesity imposes a net burden on taxpayers. "We have data on BMI and medical care costs for only a single interval of time for each respondent," Cawley and Meyerhoefer write. "Ideally we would have longitudinal data that would allow us to examine the life cycle impacts of obesity, including early mortality (the costs of which are presumably mainly internal to the household)." Early mortality, by reducing spending on Social Security and Medicare, would mitigate the fiscal impact of obesity. Such effects seem to make smoking a net money saver for taxpayers. Citing Eric Finkelstein, Reuters says obese people do not die early enough to make up for their extra yearly medical expenses. Yet a 2008 Dutch study published in PLoS Medicine found that eliminating obesity would increase total health care spending in the long run, and that does not take into account increased spending on old-age benefits such as Social Security. 

But even if Finkelstein is right, we should be wary about agreeing that the net public burden he attributes to obesity justifies a government interest in what we eat and how much exercise we get. Forcing taxpayers to subsidize other people's medical care inevitably means forcing them to subsidize other people's risky lifestyles. Just as inevitably, people will cite that reality to justify increasingly intrusive policies aimed at discouraging such lifestyles. To me, that's an additional argument against forcing taxpayers to subsidize other people's medical care, which only encourages the totalizing ambitions of public-health paternalism.

For more about the War on Fat, see my 2004 Reason cover story.